Inside the TAL incubator pursuing AI transformation

Life insurer TAL is hoping to use artificial intelligence to streamline claims processing, reduce “form filling” and get new customers on the books much faster.

The company, which is part of Japan’s Dai-Ichi Life Group, has established an internal incubator - at arm’s length from the business but directly reporting in to the CEO - to pursue its artificial intelligence and broader innovation ambitions.

TAL’s general manager of innovation Dan Taylor said the firm already has two pilot projects underway using chatbots and an application of machine learning in “soft launch”, but he sees more opportunities than he can potentially keep track of.

“We see massive potential around AI,” Taylor said.

“I’ve pretty much lost count of the different use cases we see in the value chain of the business. I can see this as being a pretty big ongoing program for us over the next few years.

This case study is part of a feature on the adoption of artificial intelligence within Australian enterprise. Download the report to read more individual case studies as well as analysis of the broader state of play.

“We’re taking the first few steps now, building our own [internal] capability, learning where the technology is most effective and some of the constraints and how to get around them, and proving the benefits.

“In the longer term, I see a world where we’ve got far simpler processes and that fundamentally means that our customers have a much better experience. That is ultimately our guiding light.”

For Taylor, this means everything his incubator touches needs to have some sort of customer benefit, whether the project be directly customer-facing or internally-focused.

“We’re very much about trying to identify what the customer challenges are that this can really improve and make a difference to, and secondly how do we reinvent our offer - whether that means changing process or product - to completely transform the customer experience,” he said.

“For example, if we can reduce the administrative burden on our staff then they can spend more time focusing on value-added services for the customer and actually helping the customer.”

Of the three pilot projects already underway through the incubator, the one in soft launch uses machine learning to improve backend processes and compliance around underwriting.

Underwriting involves assessing the risk a customer poses when they apply for insurance, whether to insure them and how much they should pay.

“When it comes to underwriting and getting people on the books, can we make that faster, simpler and easier?” Taylor said.

“At the moment it’s a slow and pretty arduous process, both on our side of the fence but also for customers.

“There’s an awful lot of forms to fill in and then bits of paper get pushed backwards and forwards between multiple people, not just between the customer and us but often advisers and doctors may also be involved.

“If we can really speed that up and make it simpler then that is going to be a massive benefit for us.”

In addition to improving the insurance application process, Taylor sees opportunities to similarly streamline the lodgement and processing of claims.

“When it comes to claims, how do we make it less about form-filling and [more about] helping you bounce back and your recovery?” Taylor said.

“When our customers claim, we [want to be able to] focus on helping them bounce back, return to work, and get the support they need, rather than filing paperwork.”

Taylor declined to reveal what technology TAL is using to back the underwriting project. Some of the algorithms were created in the incubator, albeit using “classical tools”.

TAL’s other two pilot projects involve chatbots deployed in Facebook Messenger. Taylor said both are “relatively simple” and sit atop “a rules engine”, which constrains their responses.

The company may also be able to take learnings from other parts of the Dai-Ichi group, including its corporate headquarters in Japan which is using IBM’s Watson software to assess payments.

“We’re able to understand some of the benefits they’ve seen from that and some of their experiences working with IBM,” Taylor said.

“Equally some of the projects we’re working on they’re able to learn from so I think the sharing of lessons goes both ways.

“The one caveat is their business context and regulatory environment is very different so sometimes there are things that are less relevant to each party.”

The incubator approach

One thing that sets TAL apart from others in their adoption of AI is its structure: it is effectively building up its AI capability outside of the business, though what it comes up with will affect what the business does.

Taylor believes the structure is important because it allows AI to develop in a way where it is unconstrained by “existing infrastructure and processes”.

“Our job is to be able to experiment and understand how these technologies are going to help transform our office in the future,” Taylor said.

“If this team was sat in the core business – I’ve seen this in so many organisations where this happens – they actually get drawn into this week’s problem rather than the big picture.”

Taylor said the structure meant his team of data scientists, IT, developers and operations personnel could “think more outside the box”.

The incubator also had the autonomy not just to think and strategize but also to spin up proof-of-concept and pilot projects to test its ideas.

Download the report now